In structure-based drug design, the performance of current scoring functions in binding affinity prediction is not very promising. To tackle this problem, we previously reported the Knowledge-Guided Scoring (KGS) method. The key idea is to compute the binding affinity of a given protein-ligand complex based on the known binding affinity of an appropriate reference complex. In this study, we have developed an upgraded version, i.e. KGS2, where the major improvement is to introduce 3D protein-ligand interaction fingerprints into the reference selection. KGS2 and KGS have been evaluated in combination with five selected scoring functions (X-Score, ChemPLP, ASP, GoldScore and ID-Score) on complexes formed by four established drug targets (HIV-1 protease, carbonic anhydrase, beta-secretase, and beta-trypsin). Considerable improvements were observed for HIV-1 protease, beta-secretase, and beta-trypsin complexes when KGS2 was applied, and in most cases, KGS2 was more effective than KGS by several fold. KGS2 can serve as a convenient "add-on" to current scoring functions without the need to re-engineer them, and its application is not limited to certain target proteins. Thus, it can be a very practical method for enhancing the performance of current scoring functions in binding affinity prediction.

Improved prediction of protein-ligand binding affinity on not-so-big data

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.